India has 22 official languages, but over 1,200 spoken languages. For decades, this meant a citizen’s ability to access government services often depended on whether they spoke the language of the bureaucrat behind the desk. Translation backlogs, forms in unfamiliar scripts and call centres that couldn’t understand you created a friction that excluded hundreds of millions from the digital economy.
In 2022, the government launched Bhashini, an AI-powered translation platform that is rapidly dismantling these barriers. Today, it processes over 300 million translations per month across 35 languages, embedded directly into the digital infrastructure Indians use daily. Unlike commercial tools, Bhashini is designed for public service delivery: a pensioner in Odisha can now file a grievance in Odia without navigating Hindi forms, and a farmer in Punjab can access state agricultural subsidies via voice command. At last year’s Maha Kumbh Mela – a religious gathering of tens of millions – pilgrims used the platform’s AI chatbot in 11 languages to locate lost relatives and access emergency medical services.
What decades of policy had struggled to fix, a sovereign software layer has begun to dissolve.
India is not alone. In Nigeria, fintech and AI-driven credit scoring are bridging gaps in banking infrastructure, bringing SMEs into the formal economy faster than traditional branch networks ever could. Across the Global South, governments are discovering that, beyond efficiency, AI also offers velocity: the potential to collapse decades of institutional development into years.
This matters because the terms of global competition are shifting. Since the end of the Second World War, emerging economies have competed on tangible assets: young populations, competitive labour costs, natural resources. These edges are eroding. Increasingly, a nation’s position in the global economy is defined by the friction of its digital infrastructure – its ability to verify identify, extend credit, clear customs and manage energy grids with precision. As supply chains become automated, capital will flow toward jurisdictions that can integrate with these systems. The rest risk being bypassed entirely.
The question now is how governments – especially smaller ones – access this capacity.
In 2025, the global AI race was measured in hardware: general processing units stockpiled, data centres built, benchmarks hit. Attention centred on whether OpenAI or Anthropic would reach artificial general intelligence first, and whether China was closer than Washington wanted to believe. This is the race that moved markets and consumed headlines. But for the nearly 190 countries outside the frontier-lab competition, it has been a spectator sport. Most will not build frontier AI models. But they can deploy intelligence through the systems that run their states.
This is the race most countries are actually in.
Yet the path is narrow. Unlike the jump from landlines to mobile phones, AI capacity cannot be downloaded or bought off the shelf. It requires industrial foundations – data centres, gigawatts of stable power, and land – and the institutional frameworks to govern it. This is capital-intensive nation-building, closer in complexity to electrification than to software procurement.
Emerging economies possess a distinct advantage in this moment. They are less burdened by legacy systems that trap advanced economies in expensive upgrades. And higher public trust in AI across the Global South – compared to the scepticism prevalent in advanced economies – gives governments greater political room to move and deploy new systems at speed.
But the capacity gap is stark. Excluding China and India, the Global South hosts less than 3 per cent of the world's high-performance computing capacity. This leaves many governments facing a difficult choice: partner with one of the major providers (mainly US or Chinese companies) or go without. Nations that rush to deploy without the capacity to govern the technology risk failed pilots at best, or partnerships that become dependencies when governments lack the skills to audit, adapt or eventually localise these systems.
Tech firms have recognised this gap, pitching "sovereign AI" and the “sovereign cloud” as solutions. These are necessary, but not sufficient if they address hardware without the surrounding ecosystem.
Sovereignty does not mean autarky. No country that imports food or medicine defines it that way. It means agency: the ability to audit systems, set standards and change providers without the state grinding to a halt. That requires skills, governance frameworks and competitive markets for AI services, not just servers.
This demands a comprehensive strategy that builds entire ecosystems.
Fifteen years ago, the UAE recognised this lesson in energy. When it partnered with Pacific island nations to deploy renewables, the primary barrier wasn't the technology. Solar panels were available. The barrier was the ecosystem: grid resilience, operator capability, financing structures and long-term maintenance.
Through the UAE-Pacific Partnership Fund, the UAE funded feasibility studies, grid integration frameworks and operator training. In Tonga, a utility-scale solar project demonstrated that renewables could stabilise the grid, supplying up to 70 per cent of peak demand on an island previously dependent on imported diesel. The export was not just the solar panels, but the whole blueprint for an energy transition.
The same logic is now being applied to intelligence.
Domestically, platforms like TAMM – an AI-powered government services app – in Abu Dhabi have re-engineered government processes around the population, migrating over 90 per cent of residents to digital channels. Internationally, the UAE is becoming a node in the global AI system. Through Stargate, it is positioning itself as the deployment hub for the Middle East, Africa and South Asia – regions that represent the bulk of the world's population growth and the next wave of digital demand.
The ambition extends beyond competing with frontier labs to become the bridge between them and the needs of countries’ populations, translating AI breakthroughs into systems governments can actually deploy to improve lives and grow prosperity.
This mission is critical. If we do not solve the governance and capability gap, AI's benefits will cascade upward, concentrating wealth and capability in the handful of countries and companies that already possess both. Development institutions, multilateral banks and donors must treat digital capacity with the same urgency they bring to physical infrastructure.
Most countries are not vying for supremacy in AGI. They are competing for the digital edge that will shape GDP growth, secure autonomy and attract investment.
In 2025, AI leadership was counted in chips. In 2026, the measure that matters is safe, sovereign deployment. For most countries, that race is still wide open.
